Kirana AI

What do they actually do

Kirana AI connects an on‑premise GPU appliance to a store’s existing security cameras and runs computer‑vision models continuously to spot theft events (walkouts, concealment, no‑go zone violations), safety hazards (slips/spills), and empty shelves. When something happens, it sends a short video clip and suggested action to the manager’s phone/desktop; no new cameras are required Kirana AI site YC profile.

The current workflow is: install the edge device/software, process live camera feeds 24/7, push real‑time alerts with clips and locations, and keep an audit trail so staff can intervene and follow up on incidents. The company is early (YC S25) and is focused on independent grocers and small retailers; the site also claims fast payback ("ROI in 30 days") Kirana AI site YC profile.

Near‑term, they aim to integrate with POS and ordering systems so camera signals can drive automated reordering, pricing recommendations, and other agent‑like workflows; these are roadmap items rather than fully shipped features today YC profile Kirana AI site.

Who are their target customer(s)

• Independent single‑store grocer (owner‑operator): They can’t watch the floor constantly, so theft, missed spills, and empty shelves go unnoticed and cut into already thin margins.
• Small multi‑location grocer (few stores under one owner): They struggle to supervise multiple sites and reviewing footage is time‑consuming, so shrink, inconsistent restocking, and safety issues compound.
• Store manager at a convenience or neighborhood grocery: They juggle customers, stocking, and cleaning, so urgent issues (shoplifting, spills, out‑of‑stocks) are often discovered late.
• Loss‑prevention or operations lead at a small chain: They lack budget for dedicated LP staff and need real‑time, verifiable alerts and clips instead of hours of manual video review.

How would they acquire their first 10, 50, and 100 customers

• First 10: Founder‑led on‑site pilots with nearby independent grocers: plug into existing cameras, run for a short period, deliver clips and a simple incident/ROI report to convert and secure testimonials.
• First 50: Replicate the pilot playbook with 1–2 local reps and a few channel partners (POS integrators, camera/security installers, regional wholesalers) using straightforward referral/commission terms.
• First 100: Offer a paid pilot SKU and basic reseller program with training/SLAs, add a light self‑install flow for multi‑store owners, and use early case studies to list on POS/retailer marketplaces and target regional associations.

What is the rough total addressable market

Top-down context:

India has an estimated ~12–13 million kirana/traditional grocery stores, representing the bulk of small‑format retail units globally Invest India. In the U.S., there are ~152,255 convenience stores (about 60% single‑store operators) and ~45,575 supermarkets, some portion of which are independents NACS FMI.

Bottom-up calculation:

A conservative store‑count TAM starts with India’s ~12–13M kiranas, adds ~90k U.S. single‑store c‑stores plus tens of thousands of independent groceries, and acknowledges Europe/other regions contribute tens to hundreds of thousands more—yielding a global target above 13M small stores today Invest India NACS FMI McKinsey EuroCommerce.

Assumptions:

• Stores have usable camera coverage, reliable power/networking, and can host an on‑prem GPU (or accept a provided appliance).
• Sufficient budget and willingness to pay for hardware + recurring software justify a short payback period.
• Local privacy/security rules and landlord policies permit continuous video analytics.

Who are some of their notable competitors

• Trigo: Uses store cameras plus POS to build a virtual basket and flag mismatches for loss prevention; overlaps on camera+POS loss alerts but targets larger retailers and autonomous/checkout‑heavy deployments Trigo.
• Simbe (Tally): Autonomous robots scan shelves to detect out‑of‑stocks and pricing errors; overlaps on availability monitoring but relies on dedicated mobile hardware rather than always‑on security camera feeds Simbe.
• Everseen: Vision AI focused on front‑of‑store/checkout shrink (non‑scans, product switching) with real‑time staff nudges; strongest at checkout scenarios and large retail rollouts Everseen.
• Trax: Image recognition and shelf analytics for product availability and merchandising; overlaps on shelf monitoring but is primarily a CPG/retail analytics platform rather than a small‑store, edge LP system Trax.
• Verkada: Cloud‑managed security cameras with built‑in analytics (loitering, line‑crossing, people counting); overlaps on safety/loss alerts but is a security camera platform, not an end‑to‑end retail operations manager Verkada.